What is the clinical evidence from this study?

Study design: Other. Population: Atrial Fibrillation and Parkinson's Disease. Intervention: Time-frequency time-space LSTM (TF-TS LSTM) vs. Conventional LSTM. Primary outcome: Classification accuracy for ECG atrial fibrillation and normal sinus rhythm.

What does this research mean for the field?

Integrating time-frequency and time-space properties into Long Short-Term Memory (LSTM) networks enables highly accurate and time-efficient classification of physiological signals while reducing the need for multiple wearable sensors. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

March 25, 2021Open Access

Time–frequency time–space LSTM for robust classification of physiological signals

Key Result

The proposed time-frequency time-space LSTM algorithm achieved 93.77% accuracy for ECG atrial fibrillation classification and 100% accuracy for Parkinson's disease gait classification, outperforming conventional LSTM.

Structured PICO

Population

Physiological time series data including ECG signals for atrial fibrillation and gait dynamics for Parkinson's disease used to evaluate a novel deep learning classification algorithm.

Intervention

Time-frequency time-space LSTM (TF-TS LSTM) using bi-LSTM with four extracted features (instantaneous frequency, spectral entropy, fuzzy recurrence image entropy, and fuzzy recurrence entropy).

Comparator

Conventional bi-LSTM (direct use of time series) and previously published machine learning methods.

Outcome

Classification accuracy, sensitivity, specificity, precision, F1 score, and training time.

The proposed time-frequency time-space LSTM architecture significantly improves classification accuracy and drastically reduces training time for physiological signals like ECG and gait dynamics.

Main Result

Absolute Event Rate: 93.77% vs 58.37%

Limitations

Tests of the TF-TS LSTM for the gait classification using data recorded from other single sensors were not carried out.
Difficult to fairly implement compared methods for the classification of the ECG data without the provision of the source codes.

Abstract

Automated analysis of physiological time series is utilized for many clinical applications in medicine and life sciences. Long short-term memory (LSTM) is a deep recurrent neural network architecture used for classification of time-series data. Here time-frequency and time-space properties of time series are introduced as a robust tool for LSTM processing of long sequential data in physiology. Based on classification results obtained from two databases of sensor-induced physiological signals, the proposed approach has the potential for (1) achieving very high classification accuracy, (2) saving tremendous time for data learning, and (3) being cost-effective and user-comfortable for clinical trials by reducing multiple wearable sensors for data recording.

KI fragen

Bookmark

View Full Paper

KI fragen

Bookmark

View Full Paper

Cite This Study

Tuan D. Pham (Thu,) führte eine weitere Studie zu Vorhofflimmern und Parkinson durch. Die Zeit-Frequenz-Zeit-Raum-LSTM (TF-TS LSTM) wurde hinsichtlich der Klassifizierungsgenauigkeit für das ECG bei Vorhofflimmern und normalem Sinusrhythmus bewertet. Der vorgeschlagene Zeit-Frequenz-Zeit-Raum-LSTM-Algorithmus erreichte 93,77 % Genauigkeit bei der Klassifizierung von Vorhofflimmern im ECG und 100 % Genauigkeit bei der Klassifizierung des Gangs bei Parkinson, was besser war als bei herkömmlichem LSTM.

synapsesocial.com/papers/6a26ada656d154c670c2b20f https://doi.org/https://doi.org/10.1038/s41598-021-86432-7

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: